Hand entrapment accidents that occur in industrial settings are one of the major types of accidents that seriously threaten the safety of workers, and occur particularly frequently when handling high-speed machinery or heavy equipment. Existing hand entrapment accident prevention technology is mainly based on a single camera system, which has limitations in effective prevention due to limited field of view and difficulty in accurately predicting location. This study aims to design and implement a hand entrapment accident prevention system using geometric information in a low-cost multi-camera environment. A multi-camera system collects images from multiple angles and provides three-dimensional information to accurately track the position and movement of the hand within the workspace. This allows you to calculate the distance between your hand and the danger zone in real time, and to predict and warn of potential accidents in advance. In this study, we propose an algorithm that can efficiently process real-time image frame from multiple cameras and monitor the distance between the hand and the danger zone in real time. The execution results of the implemented system showed that the average FPS based on low-cost cameras was 19.5, and when using high-performance cameras, an average of 30 FPS was achieved, successfully providing real-time warnings in case of intrusion into danger zone. Through this study, we aim to present practical and effective hand-trapping accident prevention technology that can further improve worker safety.